Greetings Qrew,

Whether you're diving into AI for work, side projects, or just out of curiosity — this issue of The Creators Hive brings you practical insights, hands-on learning, and tools you can actually use.

1. Kimi K2 Thinking takes open-source to new level
2. Microsoft establishes new Superintelligence Team
3. Apple taps Google’s Gemini for Siri overhaul
4. Ex-Meta designers launch Stream Ring AI wearable
5. Google’s space-based AI data centers
6. Tool Recommendation of the Day: Runable

---

Kimi K2 Thinking takes open-source to new level

Alibaba-backed Moonshot AI has launched Kimi K2 Thinking, an open-source reasoning model that rivals or surpasses GPT-5 and Claude 4.5 Sonnet across multiple benchmarks — all while costing a fraction of what frontier models require to train.

Key details:

• Kimi K2 Thinking outperformed GPT-5 and Sonnet 4.5 on several agentic and reasoning benchmarks, achieving a record 44.9% score on Humanity’s Last Exam.
• The model showed massive improvements in coding performance over its previous version (released just four months ago), though it still slightly trails top-tier systems.
• It can autonomously chain 200–300 tool calls to complete complex workflows, while also showing strong results in creative writing and long-context reasoning.
• Training reportedly cost under $5 million, with inference pricing significantly lower than competing frontier models.

Why it matters: Just days after Nvidia’s Jensen Huang said China was “nanoseconds behind” in AI, Moonshot’s release seems to prove the point. K2 Thinking represents a leap forward for both Chinese and open-source AI — offering frontier-level reasoning at breakthrough affordability, and signaling that the open ecosystem is closing the gap faster than expected.

---

Microsoft establishes new superintelligence Team

Microsoft AI CEO Mustafa Suleyman has announced the formation of the MAI Superintelligence Team, a new research group focused on developing “Humanist Superintelligence” — advanced AI designed to solve real-world problems in medicine, education, and energy rather than pursuing open-ended AGI.

Key details:

• Suleyman described the mission as creating AI that “always works for, and in service of, people and humanity,” signaling a values-driven approach to superintelligence.
• The team’s focus areas include medical superintelligence, AI learning companions, and breakthroughs in clean energy innovation.
• Karen Simonyan, Suleyman’s co-founder at Inflection, joins as Chief Scientist, alongside researchers recruited from DeepMind, OpenAI, and Anthropic.
• The announcement follows Microsoft’s revised deal with OpenAI, which now allows both companies to independently pursue superintelligence development.

Why it matters: This move gives Microsoft’s AI ambitions a distinct philosophical and strategic identity for the first time. Rather than chasing generalized AGI, the company is positioning itself to build purpose-driven, human-centered AI systems — signaling a shift toward pragmatic, socially aligned superintelligence development.

---

Apple taps Google’s Gemini for Siri overhaul

Apple has reportedly finalized a $1B-per-year deal with Google to integrate a custom 1.2 trillion–parameter version of Gemini into Siri, marking a major shift in Apple’s AI strategy as it works to modernize its long-stagnant assistant.

Key details:

• Gemini will power key Siri capabilities like summarization and multi-step planning, hosted on Apple’s Private Cloud Compute to ensure user data privacy.
• Apple tested models from OpenAI and Anthropic before selecting Gemini, which is significantly larger than Apple’s current 150B-parameter in-house model.
• The partnership will remain largely behind the scenes, with Google acting as a quiet infrastructure and model provider rather than a public partner.
• The upgraded Siri is expected to debut next spring, with Apple viewing Gemini as an interim solution while it continues developing its own long-term AI model.

Why it matters: After years of lagging behind in AI assistants, Apple is making a bold — and pragmatic — move by leaning on its rival’s technology. While the partnership may be temporary, it underscores how even the world’s most valuable company is willing to outsource intelligence as it races to stay competitive in the rapidly evolving AI ecosystem.

---

Ex-Meta designers launch Stream Ring AI wearable

Sandbar, a new startup founded by ex-Meta designers, unveiled the Stream Ring — an AI-powered wearable that captures whispered thoughts, transcribes them into organized notes, and doubles as a hands-free music controller.

Key details:

• Cofounders Mina Fahmi and Kirak Hong previously worked on neural interface tech at CTRL-Labs, acquired by Meta in 2019.
• The device uses whisper-detection microphones and a touchpad activator instead of voice wake words, enabling private, quiet note-taking.
• Its built-in AI assistant replies in a synthetic version of the wearer’s own voice, powered by ElevenLabs speech synthesis.
• Preorders start at $249, with an optional $10/month subscription, and delivery slated for summer 2026.

Why it matters: Stream Ring adds to the growing wave of AI wearables focused on frictionless, context-aware interaction. Its whisper-based interface and minimal design could set it apart in a space crowded with pins, pendants, and earbuds — if Sandbar can deliver on its ambitious promise of blending subtlety, utility, and personal AI.

---

Google’s space-based AI data centres

Google unveiled Project Suncatcher, a bold research initiative exploring AI data centers in space — using solar-powered satellites equipped with Google’s AI chips to process workloads in orbit and bypass Earth’s mounting energy constraints.

Key details:

• The project envisions orbiting satellites generating solar energy at up to 8x Earth efficiency, providing continuous, uninterrupted power.
• Google’s AI chips have passed radiation durability tests simulating five years in space, solving a key issue that typically disables standard processors.
• A 2027 trial is planned with two satellites in collaboration with satellite imaging firm Planet to assess real-world performance in orbit.

Why it matters: With AI infrastructure demands surging, Google is literally taking compute off-planet. If feasible, space-based AI could overcome terrestrial bottlenecks like power shortages, high cooling costs, and community resistance — potentially setting the stage for the next frontier in cloud infrastructure.

---

Tool Recommendation of the Day: Runable

Website: runable.com

What makes it unique

Runable is a no-code/low-code AI automation platform that lets you describe tasks in plain English and it handles everything—from website actions, content creation, workflows, to automations across apps—without needing to write scripts.

Key Features

• Use natural-language prompts to kick off multi‐step workflows (e.g., “Check my calendar, do company research, update Google Sheets”).
• Pre-built templates (called “runbooks”) for marketing, sales, research, data analysis, reporting, productivity, education templates.
• Integrations with calendar, Google Docs/Sheets, email, forms, and more; ability to automate tasks across systems.
• Save workflows, schedule them for recurring runs, and reuse/edit them as needed.

Use case magic

• Automatically prepare meeting briefs by pulling calendar info, researching participants, and saving notes.
• Automate social-media content creation and scheduling from a prompt.
• Generate analytical reports or dashboards from raw data with minimal manual input.
• Reduce repetitive digital tasks so you can focus on strategic or creative work.

Why it stands out

Many platforms automate one or two kinds of tasks (email, spreadsheets, scheduling). Runable aims to be a generalist automation agent: you tell it what you want done and it figures out the steps across tools. It bridges no-code accessibility with broad task scope.

---

Masterpieces from the Hive 🎨

Here’s what our community has been creating this week using the Hive Wall:

​

32 North Gould Street, Sheridan, Wyoming 82801
​Unsubscribe · Preferences​